This project aims to develop an assistive system using smart glasses to enhance the independence and navigation of visually impaired users. The system integrates object detection, voice generation, and wireless communication modules to provide real-time feedback and interaction with the environment. The object detection module utilizes the Jetson Nano and NVIDIA's Jetson Inference library to identify and classify objects within the user's surroundings, such as furniture and household items. This system processes live video feeds from a connected camera, providing audio guidance or haptic feedback to inform the user of detected objects. To optimize performance, the model was retrained to focus on common household items, enhancing precision while reducing computational load.

The voice generation module provides auditory feedback by converting text into speech, essential for users with visual impairments or those seeking a hands-free experience. This system uses the Jetson Nano and a text-to-speech engine to deliver clear auditory responses, while addressing challenges in speech quality and power consumption. Pre-generated audio files for frequent commands help mitigate processing demands.

For wireless communication, the system uses an IoT camera for live video streaming and Arduino Pico boards for controlling external devices like lights or thermostats. Wireless protocols such as MQTT and WebSocket ensure efficient transmission of data, with security measures like WPA2 encryption to protect sensitive information.

The development process has involved significant optimization for low-power, high-efficiency operation, while maintaining robust functionality. Future work will focus on further enhancing detection accuracy, refining voice quality, improving power usage, and expanding device compatibility. This integrated system demonstrates the potential to assist visually impaired individuals in navigating complex environments.